What's an Acceptable Turnitin Similarity Score?

There isn't one, and that is the honest answer. Turnitin has never published an acceptable percentage, and most universities have not bothered to either. The number students trade on forums, usually somewhere between 15% and 25%, is a habit a lot of departments drifted into. Nobody set it, and on its own it settles nothing.

That answer frustrates people, so here is the more useful version. A similarity score measures overlap, not misconduct. I have seen a 4% report end in an academic integrity meeting and a 38% report get waved through without a comment. The size of the number was not what separated them. What mattered was which text matched, and whether the student had cited it.

What the score is actually counting

Turnitin runs your submission against its index: web pages, published journals and books, and papers other students handed in before you. Anything that matches gets highlighted and counted. Your percentage is just that matched text measured against your total word count. That is the entire calculation.

Look at what that definition leaves out. Nothing in it asks whether you quoted the source correctly, whether you cited it, or whether you just wrote a sentence the same way ten thousand other people have. Texas A&M's guidance on reading a Similarity Report says so outright: "the similarity score does not necessarily mean the paper was plagiarized." The software finds overlap. A person works out what the overlap means.

The color bands, and why they mislead people

In the standard Feedback Studio inbox, each report gets a color:

• Blue: 0%, no matching text at all
• Green: 1% to 24%
• Yellow: 25% to 49%
• Orange: 50% to 74%
• Red: 75% to 100%

Students read green as "passed" and yellow as "in trouble." Neither reading holds up. The bands exist so an instructor facing ninety submissions knows which reports to open first. Some LMS integrations use different ranges altogether, which means the same paper can show green in one system and yellow in another. Go by the actual percentage and the report underneath it, never the color.

Where the 20% to 25% number came from

Workload, mostly. An instructor working through a stack of essays needs somewhere to draw a triage line, and the low twenties is where most departments end up: under it, glance at the report; over it, actually read it. Turnitin's own documentation won't name a figure, and says acceptable depends on the school, the instructor, and the assignment.

That variation is real, and it runs in both directions:

• A literature review or a methods chapter is supposed to overlap with published work. Thesis committees see 20% to 30% regularly and think nothing of it.
• A reflective essay with no sources should sit near zero. At 15%, someone will ask where the matching came from.
• A lab report using a standard protocol will match every other student who ran that protocol. Departments that assign it already know.
• A first-year composition class that caps quoted material at 10% has set its own threshold, and that instruction outranks anything you read online.

If your syllabus names a number, that number is your answer. If it doesn't, ask your instructor instead of borrowing a threshold from another department's policy.

Why 5% can be worse than 40%

Here is the part that never fits in a forum reply. Similarity is distributed across your paper, and the shape of that distribution tells you more than the total ever will.

Forty percent spread across thirty sources, all cited, mostly quoted method descriptions and a long reference list, is an ordinary research paper. Five percent concentrated in one uncited paragraph taken from a single source is misconduct, and that is the version that gets escalated.

So when your report opens, ignore the headline figure for a minute and ask three questions instead. How many sources is the matching spread across? Is one source contributing a single large block? Does every highlighted passage have a citation attached? A high score built from many small cited matches is boring. A low score built from one long uncited match is the one to go fix.

What inflates your score without any wrongdoing

Most inflated scores I see come from four places, none of which involve copying anything.

Your reference list. Every entry matches every other paper that cited the same work, because citations are formatted identically by design. A thorough bibliography can add several points on its own.

Quoted material. A properly quoted, properly cited passage still matches its source, because it is supposed to. Quotation marks do not make text invisible to the matching engine.

Common phrasing. Strings like "the results of this study suggest that" belong to nobody and appear everywhere. Individually they are trivial. Across a 6,000-word dissertation chapter they accumulate.

Your own earlier work. If you submitted a proposal that later became a chapter, Turnitin may match your new draft against your old submission sitting in the student paper repository. That is worth a conversation with your supervisor before it shows up as a surprise.

One thing that does not touch this number: the AI writing indicator. It is a separate score answering a separate question, and a paper can be flagged on one and clean on the other. If AI is the part you are worried about, the boundary between using a chatbot and plagiarizing is narrower and stranger than most syllabi explain.

How to read your score before your instructor does

The practical problem with all of this is timing. Many students never see a similarity report until after submission, when whatever it says is already on the record.

If your assignment allows resubmissions, submit early and read the first report. If your institution offers Draft Coach inside Google Docs or Word, use it. If neither is available, run the draft through an outside checker to see where your matches sit, then work through the highlighted passages one at a time. That is the workflow our own free plagiarism checker is built for: red for exact matches, yellow for reworded ones, and a real source link on each so you can tell in ten seconds whether a highlight is your cited quote or something you genuinely need to rewrite.

Then fix in this order:

1. Any highlighted passage with no citation. This is the only category that is actually urgent.
2. Any single source contributing a large share of your matching, even where cited. Heavy reliance on one source reads badly regardless of attribution.
3. Long quotes that could be paraphrased and cited instead. Quote when the exact wording matters, not to fill space.
4. Nothing else. Your reference list is not a problem, and rewriting normal academic phrasing to dodge a number is how good sentences turn into bad ones.

The number is a starting point, not a grade

If someone hands you a threshold with confidence, they are describing their own department. The honest answer to "what score is acceptable" is that acceptable depends on what your assignment is, what your instructor said, and where in your paper the matching sits.

Open the report. Read the highlights. If every one of them is cited and spread across your sources, a number in the twenties is fine. If one paragraph is doing all the matching, the percentage was never the problem.